On Wed, 13 Nov 2019 12:11:02 +0000, Areligious Republican
 declaimed the following:

>
>1. Out of interest, were there to be a 32 bit application
>running under a 64 bit OS, what actually happens when an
>interrupt comes along, andinterrupt that might have system-wide
>importance?
>

 Which OS, out of curiosity -- since Raspbian is only being built as a
32-bit OS in order to be compatible with all hardware variations.

 However... Interrupts are normally handled by the kernel, and the
kernel would be whatever the OS native width is. To my knowledge,
interrupts do not run under user/application context.
>
>2. And this is really an RPi4 consideration. Are the graphics
>controled by messages left in the postbox, and if so, the full
>grapics package, normally for the 32 vit Raspbian, would actually
>be available to 64 bit code?
>

 Now you are talking the protocol between separate processors. That
protocol defines the size of data transfers. I would expect any drivers
would have been rebuilt in such a way as to handle any size mismatches
between the two processors. If the transfer goes through memory -- well,
even a 64-bit processor can write 32/16/8-bit transfers; or it writes a
64-bit transfer and the other processor performs, say, two 32-bit transfers
to obtain the same data.

 Broadcom has been secretive about the graphics core, but I'd still
recommend trying to find the spec sheet (what a misnomer -- I've some that
run 1000 pages) for the SoC which may describe the bus widths between ARM
and graphics cores, etc.



--
 Wulfraed                 Dennis Lee Bieber         AF6VN
 wlfraed@ix.netcom.com    http://wlfraed.microdiversity.freeddns.org/

--- SoupGate-Win32 v1.05